introduction to microelectronics fabrication
Post on 06-Jul-2018
229 Views
Preview:
TRANSCRIPT
-
8/17/2019 Introduction to Microelectronics Fabrication
1/53
Chapter 1
UEEP2613
Microelectronic Fabrication
Introduction to
Microelectronic
Fabrication
-
8/17/2019 Introduction to Microelectronics Fabrication
2/53
Prepared by
Dr. Lim Soo King01 Jun 2012
-
8/17/2019 Introduction to Microelectronics Fabrication
3/53
- i -
Chapter 1 .............................................................................................. 1
Introduction to Microelectronic Fabrication ................................... 1 1.0 Introduction ................................................................................................ 1
1.1 Semiconductor Materials .......................................................................... 3 1.1.1 Elemental Semiconductor .................................................................................. 4 1.1.2 Compound Semiconductor ................................................................................. 5
1.1.3 Narrow Band-gap Semiconductor ..................................................................... 6
1.1.4 Wide Band-gap Semiconductor ......................................................................... 6
1.1.5 Oxide Semiconductor.......................................................................................... 7
1.1.6 Magnetic Semiconductor .................................................................................... 7
1.2 Semiconductor Devices .............................................................................. 7 1.2.1 PN Junction ......................................................................................................... 8
1.2.2 Bipolar Junction Transistor............................................................................. 10
1.2.3 Metal Oxide Semiconductor Field Effect Transistor ..................................... 14
1.2.4 Metal Semiconductor Field Effect Transistor ................................................ 16 1.2.5 Modulation Doping Field Effect Transistor ................................................... 18
1.3 Preparation of Facility ............................................................................. 20 1.3.1 Source of Contaminants/Particles ................................................................... 21
1.3.2 Clean Facility ..................................................................................................... 21
1.3.3 Wafer Cleaning ................................................................................................. 23 1.3.3.1 Wet Cleaning ................................................................................................................. 24 1.3.3.2 Dry Cleaning.................................................................................................................. 25 1.3.3.3 Thermal Cleaning ......................................................................................................... 26
1.3.4 Gettering ............................................................................................................ 26 1.3.4.1 Alkali Gettering ............................................................... .............................................. 27 1.3.4.2 Heavy Metal Gettering .................................................................................... ............. 28
1.4 Crystal Growth ......................................................................................... 28 1.4.1 Czochralski Crystal Growth Method .............................................................. 30
1.4.1.1 Impurity of Czochralski Process .................................................................................. 31 1.4.1.2 Concentration of Czochralski Process......................................................................... 32 1.4.1.3 Pull Rate of Czochralski Process ................................................................................. 35
1.4.2 Float-Zone Crystal Growth Method ............................................................... 38
1.4.3 Bridgman-Stockbarger Crystal Growth Method .......................................... 40
1.5 Wafer Process ........................................................................................... 40
Exercises .......................................................................................................... 43
Bibliography ................................................................................................... 46
-
8/17/2019 Introduction to Microelectronics Fabrication
4/53
- ii -
Figure 1.1: Basic tetrahedral structure .................................................................................. 2
Figure 1.2: Diamond structure .............................................................................................. 2
Figure 1.3: Wafer flat for identifying the orientation and dopant type ................................ 3
Figure 1.4: Current-voltage characteristic of a pn junction ................................................ 10
Figure 1.5: The structure of a bipolar junction transistor showing two pn junctions ......... 11 Figure 1.6: Illustration of the current components of a p+np transistor ............................. 11
Figure 1.7: Biasing an n+ pn bipolar junction transistor ..................................................... 12
Figure 1.8: Family tree of field effect transistors ............................................................... 15
Figure 1.9: A 2-D structure of an n-channel MOSFET ...................................................... 15
Figure 1.10: Characteristic curve of MOSFET .................................................................... 16
Figure 1.11: 3-D structure of a simple mesa-isolated MESFET .......................................... 17
Figure 1.12: Energy band diagram of n+-Al0.3Ga0.7As/n-GaAs heterojunction ................... 19
Figure 1.13: A schematic of a recess-gate n+-AlxGa1-xAs/GaAs MODFET ........................ 20
Figure 1.14: The number of particle and diameter of particle for various classes of
cleanliness ........................................................................................................ 22
Figure 1.15: Cleaning step and composition of RCA solution ............................................. 24
Figure 1.16: A Czochralski silicon rod puller ...................................................................... 31
Figure 1.17: Equilibrium segregation coefficients for dopant in silicon .............................. 32
Figure 1.18: Doping distribution near the solid-molten interface ........................................ 34
Figure 1.19: Freezing process occurring during Czochralski crystal growth ....................... 35
Figure 1.20: Basic float-zone crystal growth method .......................................................... 38
Figure 1.21: Float-zone crystal growth process from liquid zone at the bottom moving to
top .................................................................................................................... 40
Figure 1.22: Bridgman-Stockbarger crystal growth method ................................................ 40
Figure 1.23: Chemical mechanical polishing technique....................................................... 42
-
8/17/2019 Introduction to Microelectronics Fabrication
5/53
- 1 -
Chapter 1
Introduction to Microelectronic Fabrication
_____________________________________________
1.0 Introduction
Jack Kilby was the first person to develop miniaturized transistor circuit in
1958. It was then followed by Robert Noyce and Gordon Moore, who built first planar miniaturized transistor in 1960. Thereafter, with the aid of computer and
advancement in lithography, integrated circuit IC was fast developed into ultra
large scale integration ULSI where billions of transistors can be built in a siliconchip of 3.0cm x 3.0cm area.
Integrated circuit is built mainly on silicon due to cheap, developed
processes, and abandon on the earth crust. Beside these reasons, the byproduct
produced by the process is not toxic and can be disposed easily. There are many
other available materials that can be used to build integrated circuit. They are
gallium arsenide Gas, silicon carbide Sic, gallium nitride GaN, silicongermanium SiGe and etc. Unless, it is mentioned, this lecture shall be presented
using silicon as the building element for the integrated circuit.
An integrated circuit IC consists of a single crystalline chip or it may be
called monolithic, typically 300µm to 600µm thick and covering a surface areaof 3.0cm x 3.0cm containing both active and passive elements.
The fabrication of integrated circuit IC both MOS, bipolar devices, and
other devices involves a numbers of repeated major process steps. The
processes can be broadly classified into wafer cleaning process,
photolithography (imaging, resist-bleaching and resist development), oxidation process, etching, diffusion/ion implantation, chemical vapor deposition CVD
for thin film deposition, epitaxial silicon and polycrystalline process, physical
vapor deposition/metal deposition or evaporation/sputtering, thin film such as
silicon nitride Si3 N4, titanium nitride TiN, titanium tungsten Ti-W alloy,titanium silicide TiSi2, tungsten silicide WSi2 processes, and sintering/rapid
thermal annealing RTA.
Semiconductor such germanium, silicon, and gallium arsenide have a
lattice structure consisting of two interlocking face-centered cubic lattices, and
have total of eight atoms per unit cell. This structure can be broken down into
-
8/17/2019 Introduction to Microelectronics Fabrication
6/53
01 Introduction to Microelectronic Fabrication Preparation
- 2 -
primitive cell of tetrahedral shape shown in Fig. 1.1 and diamond structure
shown in Fig. 1.2. Diamond structure has different atom type such as gallium
arsenide is called zinc blende.
Figure 1.1: Basic tetrahedral structure
Figure 1.2: Diamond structure
The most common two types of silicon crystal orientations used to fabricate the
integrated circuit are (111) and (100) types. The selection of the type depends
on the type circuit to be built. (111) orientation is widely used for fabricating
bipolar device, whilst (100) type is mainly used to fabricate MOSFET. (100)-
type has the surface state charge 30-40% less than (111) type. Thus, it is ideal touse for MOSFET fabrication in which the device is sensitive to surface trapping
and thereby affecting the mobility of the device. Bipolar device is a current
-
8/17/2019 Introduction to Microelectronics Fabrication
7/53
01 Introduction to Microelectronic Fabrication
- 3 -
driven device. Thus, surface state charge would not have significant impact onthe speed of the device. Moreover, choosing the right orientation would assist
scribing in dicing of the silicon wafer. Figure 1.3 shows one of the common
ways to identify the orientation and dopant type. Other way to identify theorientation and dopant type is by mean of notch.
Figure 1.3: Wafer flat for identifying the orientation and dopant type
In this chapter, we discuss the semiconductor materials, semiconductor devices,
the processes of fabrication of integrated circuits including the process of silicon
crystal growth, clean room requirements, and gettering process of eliminating
contaminants.
1.1 Semiconductor Materials
Semiconductor materials can be classified into many types. They are elemental
semiconductor, compound semiconductor, narrow band-gap semiconductor,
wide band-gap semiconductor, oxide semiconductor, magnetic semiconductor,
polysilicon semiconductor, amorphous semiconductor, organic semiconductor,
low dimension semiconductor, and etc. We shall discuss the important type and
a few other types.
-
8/17/2019 Introduction to Microelectronics Fabrication
8/53
01 Introduction to Microelectronic Fabrication Preparation
- 4 -
1.1.1 Elemental Semiconductor
Silicon Si, germanium Ge, and diamond C are important group IV elemental
semiconductors. These group IV elemental materials have diamond crystalstructure. Another group IV elemental semiconductor having such a structure is
alpha tin -Sn, which is also referred as gray Sn. Other elemental structures
differing from diamond structure include group III element boron
(Rhombohedral), group V material phosphorus, and group VI materials such as
sulphur S, selenium Se, and tellurium Te.
Currently silicon Si is the most important semiconductor material used in
electronic devices. Some of the important advantages of silicon Si over other
semiconductors are:
A relative ease of passivating the surface by oxidizing in a controlledmanner forming a layer of stable native oxide that substantially reduces
the surface recombination velocity.
It has hardness for making large wafers that can be handled safely
without damaging it.
It is thermally stable up to 1,1000C that allows high-temperature
processes like diffusion, oxidation, and annealing.
It is relatively low cost due to established processes.
The basic limitations of silicon Si are the magnitude and type of its energy
band-gap. Its energy band-gap is 1.12eV. It is an indirect semiconductor that
limits the application in optoelectronics and it has relatively low carrier mobility
as compared to other semiconductor such as gallium arsenide GaAs.
Emerging materials based on Si nanostructures e.g., Si nanocrystals,
quantum wires and dots, and porous Si, and SixGe1-x layers grown on Si
substrate, appear to be promising materials in various applications. Innanostructures because of quantum confinement of carriers, it leads to increase
of electron hole wave function overlap and hence, it increases photon emission
efficiency. There is a high-energy shift toward the emission blue peak.
Porous Si can be obtained from the anodic etching of crystalline Si in
aqueous hydrofluoric acid HF. It contains a network of pores and crystallites
(microscopic crystal) with sizes in the order of several nanometers. This
material exhibits relatively efficient luminescence, which is several orders of
magnitude higher than that in crystalline Si, and it is believed to be related to
the quantum confinement effects in nanocrystalline Si.
-
8/17/2019 Introduction to Microelectronics Fabrication
9/53
01 Introduction to Microelectronic Fabrication
- 5 -
1.1.2 Compound Semiconductor
There are many compound semiconductor materials. They are usually formed
from III-V group, II-VI, IV-VI, I-III-VI2 elements. III-V group semiconductorsare GaAs, GaP, GaN, A1As, InSb, InAs, InP and etc. In general, these
crystallized materials have relatively high degree of stoichiometry (chemistry
deal with the relative quantities of reactants and products in chemical reactions).
Many of these compounds such as GaAs, InAs, InP, and indium antimonide
InSb have direct energy band-gaps and high carrier mobilities. Thus, the
common applications of these semiconductors are used to design a variety of
optoelectronic devices for both the detection and generation of electromagnetic
radiation, and also in high-speed electronic devices. The energy band-gaps of
these compounds are useful for optoelectronic applications. The energy band-gap ranges from 0.17eV for InSb to 3.44eV for GaN covering the wavelength
range from about 7.29 to 0.36m, which is from infrared through visible and to
ultraviolet spectral ranges. Materials such as GaAs and InP are also extensively
used as substrates for a wide variety of electronic and optoelectronic devices
such as light-emitting devices.
II-VI compound semiconductor such as Zn and Cd-chalcogenides such as
compounds with oxygen O, S, Se, and tellurium Te cover a wide range of
electronic and optical properties due to the wide variations in their energy band-gap. These compounds are also relatively easily miscible (can be mixed well in
any proportion), which allows a continuous “engineering” of various properties.However, the preparation of high-quality materials and the processing
technologies are not sufficiently developed in comparison with those related to
silicon Si and some III-V compounds. The II-VI compounds are typically n-type
as grown, except ZnTe, which is p-type. Among these compounds, the
conductivity type of CdTe can be changed by doping, and thus n- and p-typematerials can be obtained. Others II-VI compound such as ZnSe, ZnS and CdS
can be doped to produce a small majority of holes. For device applications, it is
possible to form heterojunctions in which the n- and p-sides of the junction are
of different II-VI compound semiconductors, and to use metal-semiconductor
and metal-insulator-semiconductor structures for carrier-injection deviceapplications. All the II-VI compound semiconductors have direct energy band-
gaps, thus, high efficient emission or absorption of electromagnetic radiation
can be expected from these materials. Therefore, these semiconductors are
important mainly for their optical properties. In addition to the binary II-VI
compounds, materials such as ternary compound like Zn1-xCdxS and ZnSxSe1-x,and quaternary compound such as Zn1-xCdxSySe1-y alloys with “engineered”
properties are also of interest.
-
8/17/2019 Introduction to Microelectronics Fabrication
10/53
01 Introduction to Microelectronic Fabrication Preparation
- 6 -
IV-VI compound semiconductor like lead chalcogenides such as PbS,
PbSe, and PbTe are characterized by narrow energy gaps, high carrier
mobilities, and high dielectric constants. The unique feature of the direct energy
gap in these compounds is that its energy band-gap increases with increasingtemperature, which means the energy gap has a positive temperature coefficient,
PTC. In contrast to the temperature behavior of the energy band-gap in other
elemental and compound semiconductors, they have a negative temperaturecoefficient. Main applications of these compounds are in light-emitting devices
and detectors in the infrared spectral region.
I-III-VI2 chalcopyrite compound semiconductor such as CuAlS2, CuGaS2,
and CuInSe2 are direct semiconductors that have energy band-gaps between
1.0eV to 3.5eV. In additional CuAlS2, CuGaS2 can be made into p-type which issuitable for making heterojunction with wide energy band-gap n-type II-VI
compound semiconductor. Some possible applications of this compound
semiconductor are light emitting device and photovoltaic solar cells.
1.1.3 Narrow Band-gap Semiconductor
Narrow band-gap semiconductors such as InSb, InAs, PbSe have the energy band-gap below about 0.5eV. Such semiconductors are extensively employed in
such infrared optoelectronic device applications as detectors and diode lasers.
Photoconductive lead sulphide PbS and lead selenide PbSe detectors can be
employed in the spectral range between about 1.0 and 6.0m. Another
important material used as a detector in the infrared range is Hg1-xCdxTe.
1.1.4 Wide Band-gap Semiconductor
Wide band-gap semiconductor is also referred as refractory semiconductors
since they are employed in high temperature application. The typical types ofthis semiconductor are SiC and II-V nitrides that have high thermal
conductivity, high saturation electron drift velocity, high breakdown electricfield, and superior chemical and physical stability. The semiconductor has high
thermal conductivity indicates it can be used in high temperature at high power
level operation. It has wide band-gap that enables detection and emission of
light in short-wavelength region likes blue and ultraviolet. It has high saturation
electron drift velocity that can be used in radio frequency RF and microwave
operations. High breakdown electric field enables the realization of high power
electronic devices and also allows high device packing density for integrated
circuit.
-
8/17/2019 Introduction to Microelectronics Fabrication
11/53
01 Introduction to Microelectronic Fabrication
- 7 -
1.1.5 Oxide Semiconductor
Oxide semiconductors are also referred as semiconductor ceramics. These
materials are polycrystalline and polyphase materials with grain sizes in therange between 1.0 to 10.0m. The properties of grains and grain boundaries
play a crucial role in both the understanding and application of the materials.
Some examples of oxide semiconductors are Cu2O (2.1eV), Bi2O (2.8eV), ZnO
(3.4eV) etc. They are used in electronic devices and sensors such as positivetemperature coefficient PTC thermistor, varistor - resistor with non-linear but
symmetric current-voltage characteristics, capacitor of high dielectric constant,gas sensor, and electro-optic modulators.
1.1.6 Magnetic Semiconductor
Semiconductor compound contains magnetic ions such as Cr, Mn, iron Fe, Co,nickel Ni, and europium Eu may exhibit magnetic properties. Some oxides such
as FeO and NiO exhibit anti-ferromagnetic properties and oxide such as
europium oxide EuO and EuS are ferromagnetic properties. The semiconductor
exhibits large magneto-optical effect that can be used to design optical
modulators.
There are many other semiconductor types such as amorphoussemiconductor and organic semiconductor that are not discussed. Student may
take your own initiative to study on your own.
1.2 Semiconductor Devices
Semiconductor devices are electronic components that exploit the electronic
properties of semiconductor materials, principally silicon, germanium, and
gallium arsenide, organic semiconductors, and other semiconductors.
Semiconductor devices have replaced thermionic devices (vacuum tubes) in
most applications. They use electronic conduction in the solid state as opposedto the gaseous state or thermionic emission in a high vacuum.
Semiconductor devices are manufactured both as single discrete device and
as integrated circuits ICs, which consist of a number from a few devices to billions of devices manufactured and interconnected on a single semiconductor
substrate.
Semiconductor materials are so useful because their behavior can be easily
manipulated by the addition of impurities. Semiconductor conductivity can be
-
8/17/2019 Introduction to Microelectronics Fabrication
12/53
01 Introduction to Microelectronic Fabrication Preparation
- 8 -
controlled by introducing electric or magnetic field, by exposure to light or heat,
or by mechanical deformation of a doped monocrystalline grid. Current
conduction in a semiconductor occurs via mobile or free electrons and holes,
collectively known as charge carriers. Doping a semiconductor with a smallamount of impurity atoms, such as phosphorus or boron, greatly increases the
number of free electrons or holes within the semiconductor. When a doped
semiconductor contains excess holes it is called p-type, and when it contains
excess free electrons it is known as n-type. The semiconductor material used indevices is doped under highly controlled conditions in a fabrication facility to
precisely control the location and concentration of p-type and n-type dopants.
Many semiconductor devices had been invented in last 60-70 years since
the day first transistor was built in Bell laboratory in 1947. We shall spendsometimes in this section briefly describe some common semiconductor devicesused today. They are pn junction or diode, bipolar junction transistor, MOSFET,
MESFET transistor, and MODFET transistor.
1.2.1 PN Junction
PN junction is of great importance both in modern electronic application and inthe understanding of other semiconductor device. It is the simplest form of
homojunction formed by doping p-type impurity and n-type impurity into asingle crystal intrinsic semiconductor. It shows a very interesting behavior of
electrons and holes in the junction. William Shockley was the first person
established the basic theory of current voltage characteristics of pn junction andthis had served as the foundation for establishing other semiconductor theories.
pn junction conducts high current in one direction and conduct very small
amount of current in the reversed direction. Thus, pn junction has the property
of rectification.
The total charge of a pn junction is given by equation (1.1).
2/1
DA
ADR biS
) N N(
N N)VV(q2AQ
(1.1)
If abrupt junction is assumed at the n-type region and ND >> NA then (NA + ND)
is ND. The capacitance of the pn junction is given by equation (1.2).
2/1
R bi
AS
)VV(2
NqC
j
(1.2)
-
8/17/2019 Introduction to Microelectronics Fabrication
13/53
01 Introduction to Microelectronic Fabrication
- 9 -
where V bi is the built-in potential and is equal to V bi =
2
i
DA
n
N Nln
q
kT.
If equation (1.2) is rearranged, it becomes
AS
R bi
2
Nq
)VV(2
C
1
j
(1.3)
From the graph of 1/C j2 versus applied voltage VR , the concentration of impurity
ND and NA can be obtained from the gradient of the plot and from the
intersection at applied voltage axis by the gradient, the built-in potential V bi can be obtained. The capacitance of the junction can be measured by applying a few
milli-volt of ac signal with in frequency in 100kHz range, riding on a sweepingdc voltage to one end of the junction. The output at the other junction is dc
filtered and fed into the lock-in amplifier to measure the capacitive component
and conductive component.
The ideal current-voltage current density of a pn junction is governed byequation (1.4).
J =
1
kT
qVexp
L
nqD
L
pqDR
n
pon
p
no p (1.4)
where the reverse saturation current density is JS =
n
pon
p
no p
L
nqD
L
pqD, where D p,
Dn, pno, and n po are the diffusion coefficient of minority hole, diffusion
coefficient of minority electron, minority hole in n-region, and minority electron
in p-region at equilibrium.
Equation (1.4) becomes equation (1.5) after replacing
n
pon
p
no p
L
nqD
L
pqD
with Js.
J =
1
kT
qVexpJ R S (1.5)
-
8/17/2019 Introduction to Microelectronics Fabrication
14/53
01 Introduction to Microelectronic Fabrication Preparation
- 10 -
When the pn junction is under reversed bias, the applied voltage becomes - VR .
Expression
kT
qVexp R in equation (1.5) shall be approaching zero for |VR | >>
kT/q. Thus J = - JS when the pn junction is under reverse bias and currentdensity is independent of the biased voltage to a certain extend. Figure 1.4
shows the current density J versus applied voltage VR of pn junction.
Figure 1.4: Current-voltage characteristic of a pn junction
1.2.2 Bipolar Junction Transistor
Bipolar junction transistor BJT was invented in 1948 by John Bardeen and
Walter Brittain. The principle of operation was explained by W. Shockley ayear later. Bipolar junction transistor is a three terminal semiconductor device
used primarily for signal amplification and as switching device. It is also formed
the fundamental element for integrated circuit design such as the VLSImicroprocessor.
Bipolar junction transistor can be viewed as two pn junctions connected
back to back to form np- pn or pn-np structures name as npn or pnp transistor.
Figure 1.5 shows a structure of a bipolar junction transistor showing presence of
two pn junctions.
-
8/17/2019 Introduction to Microelectronics Fabrication
15/53
01 Introduction to Microelectronic Fabrication
- 11 -
Figure 1.5: The structure of a bipolar junction transistor showing two pn junctions
Like pn junction the current components of bipolar junction transistor come
from two carrier types, which are the hole and electron current. This is also the
reason why it is called bipolar device.
The current components shown in Fig. 1.6 are diffusion hole, diffusion
electron, drift hole, and drift electron, which already reviewed in theory of pn junction.
Figure 1.6: Illustration of the current components of a p+np transistor
Bipolar junction transistor has three terminals. One terminal is used to injectcarrier name as emitter E, one is used to control the passage of the carrier
named as base B, and one is used to collect the carrier named as collector C.
Bipolar junction transistor is designed in such that the doping concentration
of its emitter is higher than the doping concentration of the base and collector.
The order of doping concentration is highest for emitter 1018
cm-3
, followed by
collector 10x17cm-3 and than base 1016cm-3. This is to ensure closed to 100%
-
8/17/2019 Introduction to Microelectronics Fabrication
16/53
01 Introduction to Microelectronic Fabrication Preparation
- 12 -
of the injected carrier are collected by collector. The diffusion carriers of emitter
have to outnumber the recombination of carrier in the base.
The base is also designed to be much shorter than the diffusion length L p, orLn of the minority hole or electron carriers.
In normal operation of bipolar junction transistor, the emitter-to-base
junction of the bipolar junction transistor is always forward biased. Thecollector-to-base is always reverse-biased which is shown in Fig. 1.7 for n+ pn
bipolar junction transistor.
Biasing emitter-to-base of bipolar junction transistor will inject majority
electron carrier into the p-base. Some of the electrons will recombine with themajority hole carriers in the base to form part of the base current IB. Most of theminority electron carrier will reach the depletion edge in the collector and being
swept to form collector current IC. Since the injected minority electron carriers
are due to emitter current, therefore the collector current IC is BIEn + IGen + ICn +
ICp, which is approximately equal to equation (1.6) since generation current IGen
and drift hole current ICp and drift electron current ICn at collector are very small.
IC = BIEn (1.6)
where IEn is the electron current of the emitter. The factor B is called the base
transport factor and its value is less than one. The emitter current IE is made upof two components, which are injected electron IEn from n
+ to p-region and hole
injected current IEp from p to n+ region.
Figure 1.7: Biasing an n+ pn bipolar junction transistor
Thus, we can define the efficiency e of the emitter as
-
8/17/2019 Introduction to Microelectronics Fabrication
17/53
01 Introduction to Microelectronic Fabrication
- 13 -
e =I
I I
En
En Ep (1.7)
For efficient bipolar junction transistor both B and e, values must be closed tounity. The ratio of collector current IC and emitter current IE shall be
e
n b
2'
Be
EpEn
En
E
C
D2
W1B
II
BI
I
I (1.8)
where called the current transfer ratio.
The ratio of collector current IC and base current IB shall be
I
I
BI
B I I
BI
I I BI
BI
I BI
C
B
En
En Ep
En
En Ep En
En
E En
( )1
B I I
B I I
B
B
En E
En E
e
e
( / )
( / )1 1 1
= (1.9)
where is called the base to collector current amplification factor . The factor
can be quite large for BJT.
The collector current IC is approximated from equation (1.10).
IC t
b pn p
qWAnqAn
v (1.10)
where A is the cross sectional area of the device, n p is the minority
concentration in p-base, which is the injected electron concentration from
emitter into base. t is the transit time of electron to pass through the base region
of thickness W b. vn is the electron velocity through the base, which is also equal
to W b/t.
The base current IB is primarily due to recombination of injected electron
with majority carrier of the base. Thus,
B
p b
B
nAqWI
(1.11)
where B is the diffusion time of the injected electron with the hole in the base.
Combining equation (1.10) and (1.11) will yield
-
8/17/2019 Introduction to Microelectronics Fabrication
18/53
01 Introduction to Microelectronic Fabrication Preparation
- 14 -
I
I
C
B
B
t
(1.12)
Thus, a high current gain entails a low recombination rate, which means a longdiffusion time B. For indirect semiconductor material like silicon and
germanium, their recombination time is in microsecond. For direct
semiconductor such as GaAs and InGaAs, its recombination time is in the rangeof nanosecond. For high current gain, the material needs a very short transit
time t. Of course, this is would benefit the speed of the device.
1.2.3 Metal Oxide Semiconductor Field Effect Transistor
The tremendous advancement of silicon technology developed in last fifty yearshas given advantage edge to fabricate the electronic devices. This is also due tothe presence of a high quality silicon dioxide, which can be formed on silicon. It
has high degree of perfection that silicon has overtaken germanium, which was
the initial choice for transistors. Also due to perfection of Si-SiO2 interface
system that gives the reason the potential effect transistor PET like bipolar
junction transistor has been replaced by field effect transistor FET in many
applications in last four decades.
The MOSFET is the main member of the family of field-effect transistors.A transistor in general is a three-terminal device where the channel resistance
between two of the contacts is controlled by the third terminal, which is the gate.
MOSFET also has fourth terminal as contact to the substrate. The substrate isusually biased with either VDD voltage or VSS voltage depending on the type of
substrate.
A family tree of field-effect transistors is shown in Fig. 1.8. The three first-
level main members are insulated-gate field effect transistor IGFET, junction
field effect transistor JFET and metal-semiconductor field effect transistorMESFET. They are distinguished by the way the gate capacitor is formed. In an
IGFET, the gate capacitor is an insulator. In a JFET or a MESFET, the capacitor
is formed by the depletion layer of a pn junction or a schottky barrier
respectively. In the branch of IGFET, it can be further divided intoMOSFET/MISFET and heterojunction field effect transistor HFET. In the
HFET branch, the gate material is a high band-gap semiconductor layer grown
as a heterojunction, which acts as an insulator. Although MOSFETs have been
made with various semiconductors such as Ge, Si, and GaAs, and use various
oxides and insulators such as SiO2, silicon nitride Si3 N4, and A12O3, the most-
important system is still the SO2-Si interface system.
-
8/17/2019 Introduction to Microelectronics Fabrication
19/53
01 Introduction to Microelectronic Fabrication
- 15 -
Figure 1.8: Family tree of field effect transistors
A MOSFET essentially consists of a MOS capacitor and two diffused orimplanted regions that serve as ohmic contacts to an inversion layer of free
charge carriers with the semiconductor-silicon dioxide interface. Figure 1.9
illustrates the 2-D structure of an n-channel MOSFET.
Figure 1.9: A 2-D structure of an n-channel MOSFET
The linear current of the MOSFET is governed by equation (1.13).
2
VV)VV(
L
CWI
2
DSDStGS
oxnDS (1.13)
One can see that the linear current is dependent on the aspect ratio of the device,
which is dimensional dependence.
-
8/17/2019 Introduction to Microelectronics Fabrication
20/53
01 Introduction to Microelectronic Fabrication Preparation
- 16 -
After pinch-off, IDS is assumed to be constant. The VDSSAT is equal to (VGS-
Vt). Thus, the current is governed by equation (1.14).
2
)VV(VV
L
CWI
2tGS2
tGSoxn
DSSAT (1.14)
= 2tGSoxn )VV(
L2
CW
This is the equation for the saturation region of the MOSFET characteristics.
A typical ideal characteristic curve of an n-channel MOSFET is shown inFig. 1.10. The curve shows three regions of the characteristic, which are the
linear, saturation, and cutoff regions.
Figure 1.10: Characteristic curve of MOSFET
1.2.4 Metal Semiconductor Field Effect Transistor
The 3-D structure of a typical mesa isolated gallium arsenide GaAs metal
semiconductor field effect transistor MESFET is shown in Fig. 1.11. The
-
8/17/2019 Introduction to Microelectronics Fabrication
21/53
01 Introduction to Microelectronic Fabrication
- 17 -
internal pinch off voltage VP is equal to (V bi - VG), which is also called intrinsic
pinch off voltage. It is defined as
S
2
DP
2hqNV
(1.15)
where h is the thickness of the channel. The gate voltage VG required to cause pinch off is denoted by threshold voltage V t, which is when gate voltage VG is
equal to Vt. i.e. Vt = (V bi - V p). If V bi > V p, then the n-channel is already
depleted. It requires a positive gate voltage to enhance the channel. If V bi < V p,
then the n-channel requires a negative gate voltage to deplete.
The gate voltage VG needed for pinch off for the n-channel MESFETdevice is
Vt = V bi -V p=
D
C b
N
Nln
q
kT
S
2
D
2
hqN
(1.16)
where b is schottky barrier potential, which is defined as sm b . m and s
are metal work function and electron affinity of semiconductor. NC is the
effective density of state in conductor band of the semiconductor respectively.
For GaAs semiconductor, the value of NC is 4.7x1017cm-3.
Figure 1.11: 3-D structure of a simple mesa-isolated MESFET
Like the MOSFET device, the current characteristics of the MESFET have the
linear and saturation values, which are governed by the equation (1.17) and
(1.18) respectively.
-
8/17/2019 Introduction to Microelectronics Fabrication
22/53
01 Introduction to Microelectronic Fabrication Preparation
- 18 -
2/1
S
2
D
2/3
G bi
2/3
G biDD
DnD
)2/hqN(3
VVVVV2V
L
Wh NqI (1.17)
for 0 VD VDsat and VP VG 0.
I g
VV V
V V
VDsat o p
bi G
bi G
p
3
2
3
3 2
1 2
/
/ (1.18)
for VD VDsat and VG VP. go is the channel conductance, which is defined as
L
Wh Nqg Dno
.
1.2.5 Modulation Doping Field Effect Transistor
In order to maintain high transconductance for MESFET devices, the channel
conductance must be as high as possible, which can be seen from equation
(1.17) and (1.18) for MESFET device. The channel conductance is dependent
on the mobility and doping concentration. But increasing doping concentration
would lead to degradation of mobility due to scattering effect from ionizeddopant. Thus, the ingredient is to keep concentration low and at the same time
maintaining high conductivity. As the result of this need, heterojunction
modulated doping field effect transistor MODFET is the choice.
The most-common heterojunctions for the MODFETs are formed fromAlGaAs/GaAs, AlGaAs/InGaAs, InAlAs/InGaAs, and AlxGa1-x N/GaN
heterojunctions. The better MODFET is fabricated with MBE or MOCVD etc
and it is an epitaxial grown heterojunction structures.
AlxGa1-xAs/GaAs MODFET is an unstrained type of heterojunction. This is
because the lattice constants of GaAs (5.65o
A ) and AlAs (5.66o
A ) are almost the
same except the energy band-gap. The energy band-gap of gallium arsenide
GaAs is 1.42eV, while the energy band-gap of aluminum arsenide AlAs is
2.16eV. The energy band-gap of the alloy can be calculated using equation AlloyGE
= a + bx + Cx2, where a, b, and c are constant for a particular type of alloy. For
AlxGa1-xAs, a is equal to 1.424, b is equal to 1.247, and c is equal to 0.
For MODFET fabricated with AlxGa1-xAs/GaAs material, the approach is
to create a thin undoped well such as GaAs bounded by wider band-gap
modulated doped barrier AlGaAs. The purpose is to suppress impurity
-
8/17/2019 Introduction to Microelectronics Fabrication
23/53
01 Introduction to Microelectronic Fabrication
- 19 -
scattering. When electrons from doped AlGaAs barrier fall into the GaAs, they become trapped electrons. Since the donors are in AlGaAs layer not in intrinsic
GaAs layer, there is no impurity scattering in the well. At low temperature the
photon scattering due to lattice is much reduced, the mobility is drasticallyincreased. The electron is well is below the donor level of the wide band-gap
material. Thus, there is no freeze out problem. This approach is called
modulation doping . If a MESFET is constructed with the channel along the
GaAs well, the advantage would be reduced scattering, high mobility, and no
free out problem. Thus, high carrier density can be maintained at low
temperature and of course low noise. These features are especially good for
deep space reception. This device is called modulation doped field effect
transistor MODFET and also called high electron mobility transistor HEMT or
selective doped HT. Figure 1.9 illustrates the energy band diagram of n
+
-AlxGa1-xAs and n-GaAs heterojunction showing EC and EG. The delta energy band-
gap between the wide band-gap and narrow energy band-gap device are
determined from equation (1.19) and (1.20) respectively.
EC = q(narrow - wide) (1.19)
and
EV = EG -EC (1.20)
wide and narrow are respectively the electron affinity of wide band-gap and
narrow band-gap semiconductor respectively.
Figure 1.12: Energy band diagram of n+-Al0.3Ga0.7As/n-GaAs heterojunction
-
8/17/2019 Introduction to Microelectronics Fabrication
24/53
01 Introduction to Microelectronic Fabrication Preparation
- 20 -
The construction of a recess-gate AlGaAs/GaAs MODFET is shown in Fig.
1.13. The dotted line shows the quantum well where two-dimensional electron
gas 2-DEG flows. The undoped AlGaAs, which acts as buffer is 30 – 60o
A
thick. The n-AlGaAs is around 300o
A thick with concentration of approximately
2x1018
cm-3
. For recess-gate type, its thickness is about 500o
A . The source anddrain contacts are made of alloy containing germanium such as AuGe. The gate
materials can be from titanium Ti, molybdelum Mo, tungsten silicide WSi2, W
and Al.
Figure 1.13: A schematic of a recess-gate n+-AlxGa1-xAs/GaAs MODFET
1.3 Preparation of Facility
In the modern sub-micron integrated circuit fabrication, it requires a multi-
million dollar facility that consists of equipment for various fabrication process
steps, cleaning stations, and source materials. Beside the equipment, stations,materials are people who are the working in the facility.
The primary quality requirement for the integrated circuit IC facility is
cleanliness. The facility is subjected to too many sources of contaminant, which
are harmful to device under fabrication. It is the known fact that anycontaminant has a size large enough to cover the active area of a sub-microndevice. If such type contaminant is resided on the active area of device during
fabrication, the consequence is malfunction of the device. Thus, it is necessary
for a modern integrated circuit fabrication facility to keep the contaminant and
particle levels below part per million ppm or part per billion ppb level.
In modern integrated circuit facility set-up, it employs three-tiered
approaches to control the particle level and contaminant level. They are clean
facility, wafer cleaning, and gettering. We will describe these approaches in the
process of getting clean facility and gettering for integrated circuit fabrication.
-
8/17/2019 Introduction to Microelectronics Fabrication
25/53
01 Introduction to Microelectronic Fabrication
- 21 -
1.3.1 Source of Contaminants/Particles
Contaminants/particles are mainly come from people who work in the
fabrication facility, equipment in the facility, air circulating in the facility, andthe supplied materials from external vendor.
Particle contaminant such as dust particle, particle from cosmetic, powder
in the air always present in a distribution of size and shape. However, the most
concern size is between 10nm and 10m. Particle of size 10nm tends to
coagulate into a larger size. If the size is larger than 10m, it will fall by gravity
on the surface. Particle of size between 10nm and 10m remains suspend in air
for a long time. Such particle type can be deposited on the surface primary
through two mechanisms, which are Brownian motion and gravitationalsedimentation. The Brownian motion is a random motion that can occasionally
bring the particle and deposit on the surface.
People typically emit 5 to10 million particles and contaminants per minute
from each centimeter square surface. The particle/contaminant emitted by
people are mainly from exhaled air, skin, and hair etc, which consist of dust
particle, water, salt, carbon dioxide, oxygen, and contaminants such as nickel,
manganese, phosphorus etc from powder and cosmetic.
Raw materials such as acid, solvent that brought in from vendor normally
contain contaminants after handling even though they are electronic grade
materials. The contaminants are required to be filtered out before using.
1.3.2 Clean Facility
Knowing the sources of contaminant, certainly the solution is to eliminate them.If it can be eliminated, it is the best. Otherwise, controlling the amount to the
acceptable level is good enough.
A modern sub-micron IC fabrication facility normally is built with class 10or class 100 cleanliness standards. A class X simply means that each cubic foot
of air in the facility has less than X total number of particles of size greater than
0.5m. In the critical process area such as ion implantation, class 1 cleanliness
is required.
Figure 1.14 shows the number of particle versus the diameter of particle
expectation for different class of cleanliness for fabrication of the VLSIintegrated circuit. Take for an example, the class 100 environment, in one cubic
-
8/17/2019 Introduction to Microelectronics Fabrication
26/53
01 Introduction to Microelectronic Fabrication Preparation
- 22 -
foot of air, the number of particle of size greater than 0.5m should not be more
than 100.
Figure 1.14: The number of particle and diameter of particle for various classes of
cleanliness
Since people working in the integrated circuit fabrication area are continuously
emitting contaminant particles which mean this source of contaminant cannot beeliminated. As the result, the particle level in the air will be increased. Thus, the
control procedure becomes necessary.
The air in the IC fabrication facility is sucked into the air duct via the vents
mounted either on the wall or on the raise floor of the facility. The air is then
channel to the ceiling with portion of it is released to the atmosphere, portion ofit is mixed with external filter air and is forced through the high efficiency
particulate air HEPA filter residing in the ceiling at the velocity of 50cms-1
before it is released into the facility through vent mounted on the ceiling. Therelease and mixing is necessary to maintain the level of oxygen in the facility.
The HEPA filter is composed of thin porous sheets of trafine glass fiber of
diameter less than 1m. Large particles having diameter greater than 1m are
trapped by the filter, while the small particles that can pass through will be stuck
to the filter due to electrostatic charge. Even if the small particles are not
charged, due to work function difference between the particles and filtermaterials, eventually they are stuck in the filter. The air after filtered by HEPA
filter normally has cleanliness better than class 1.
-
8/17/2019 Introduction to Microelectronics Fabrication
27/53
01 Introduction to Microelectronic Fabrication
- 23 -
People working in the IC fabrication facility required to wear the “bunnysuit” and a pair of clean room shoes, which covers the body and cloth to block
the source of contaminant from fabric, human hair, and human sweat. The people are also required to wear face mask to block the particles from the
exhaled air getting into the facility. Before entering into the fabrication facility, people have to be cleaned by air shower. Air shower will blow away loose
particles/contaminant residing on the cloth and body. Beside all these
procedures, people who are working in the fabrication facility are barred from
using cosmetic and powder.
Chemical such as sulfuric acid, hydrogen peroxide, acetone, aqueous
alkaline, etc used for fabrication process are to be specified as electronic gradetypes with level of contaminant control to a specific electronic grade standard.
Water is an important solvent for cleaning and rinsing purposes. City water
from the tap is too dirty containing too much of contaminants and particles suchas chlorine, heavy metals, silt, bacteria etc and is not suitable for cleaning and
rinsing. De-ionized DI water is normally used. DI water is the highly purified
and filtered water obtained from reverse osmosis process through 0.1nm filter. It
has all traces of ionic, particles, and bacterial contaminants been removed.
Another important parameter of DI water is the resistivity. This parameter
is important because too low the resistivity value means too much the
dissociation of water molecule i.e. too much H+ and OH
- ions, which would
create too much static charge problem on wafer knowing that static charge will
attract particles and contaminants.
A basic DI water system can achieve resistivity of 18.0Mohm-cm versus
the theoretical resistivity of pure water at temperature 250C, which is
18.3Mohm-cm, with fewer than 1.2 colonies of bacteria per milliliter and no
particle size larger than 0.25m.
1.3.3 Wafer Cleaning
By nature, there is a layer of native oxide grown on silicon wafer due to
presence of oxygen in the atmosphere and also due to presence of contaminants
such as wax, resin, greasy film, sodium chloride, copper, and etc. Moreover,
each process steps, inorganic and organic chemicals such as organic photoresist,
hydrochloric acid, developing solution etc are used. The residue of the chemical
has to be cleaned before proceeding to next process step. Thus, it is necessary to
-
8/17/2019 Introduction to Microelectronics Fabrication
28/53
01 Introduction to Microelectronic Fabrication Preparation
- 24 -
clean the wafer before proceed to next fabrication steps. Today there are two
types of wafer cleaning technologies namely the wet clean and dry cean
technology.
1.3.3.1 Wet Cleaning
There are two types cleaning solution for wet cleaning, one for removing
organic material and one for removing inorganic material.
Hydrofluoric acid is normally used to remove oxide that formed on surface
of silicon wafer. Ammonium hydroxide, sulfuric acid, and hydrogen peroxide
are the chemicals typically used to remove organic contaminants, whilst
hydrogen peroxide and hydrochloric acid are chemicals used to removemetal/inorganic contaminants. De-ionized DI water is generally used as solventfor cleaning or rinsing. The wafer is finally dried in nitrogen environment and
keeps in the storage cabinet with nitrogen circulation to prevent oxidation and
contamination.
The right proportional mixture of the above mentioned solvents are termedas Radio Corporation of America RCA solution that was developed in 1965.
The solutions are divided into solution clean 1 and solution clean 2. Figure 1.15
shows the eight cleaning steps for cleaning the wafer to remove inorganic,
organic, and native oxide contaminants before actual fabrication process steps
begin. The figure also shows the composition of various solutions andtemperature requirements during cleaning process.
Step Solution Temperature Type of Contaminant to be removed
1 H2SO4 + H2O2 (4:1) 1200C Organic particle
2 DI water 25 C Rinse
3 NH3OH + H2O2 + H2O (1:1:5)RCA clean 1 80
0
C – 900
C Organic particle
4 DI water 25 C Rinse
5 HCl + H2O2 + H2O (1:1:6)
RCA clean 280
0C – 90
0C Inorganic ion
6 DI water 25 C Rinse7 HF + H2O (1:50) 25
0C Native oxide
8 DI water 25 C RinseFigure 1.15: Cleaning step and composition of RCA solution
-
8/17/2019 Introduction to Microelectronics Fabrication
29/53
01 Introduction to Microelectronic Fabrication
- 25 -
1.3.3.2 Dry Cleaning
Wet cleaning method still remains as the major method for wafer cleaning.However, in ULSI microelectronic industry, it has issue related with particle
generation, dry difficulty, cost, chemical waste disposal, general flexibility etc.
Dry cleaning method provides an alternative for cleaning wafer. Dry cleaning is
a gas phase chemistry that requires excitation energy to enhance the chemicalreaction at low temperature. The added energy can be plasma, particle beam,
short wavelength radiation, or thermal cleaning.
A number of dry cleaning methods have been developed namely HF/H 2O
vapor cleaning, ultraviolet-ozone cleaning, H2/Ar plasma cleaning, and thermalcleaning. These clean methods are usually used after wet cleaning. Dilute HFacid in wet cleaning usually creates hydrocarbon contamination on the surface
of wafer. However, with HF/H2O vapor dry cleaning, the wafer surface is
fluorine terminated.
Ultraviolet-ozone cleaning UVOC is an effective way to removehydrocarbon, although the surface is oxide passivated. The chemical process of
UVOC can be explained from equation (1.21) to (1.24). Equation (1.21) is a
surface excitation process.
Absorbed impurity + hv o
AUV30002000 Excited impurity (1.21)
In addition, oxygen molecules are excited to UV light to form atomic oxygen.
The gas-phase excitation processes are shown in the following equations.
O2hOo
AUV1849
2 (1.22)
32 OOO (1.23)
2
AUV2537
3 OOhOo
(1.24)
Then the excited impurity like hydrocarbon would react with oxygen atom and
ozone to form volatile compound.
Hydrofluoric acid in RCA clean1 and clean 2 solutions promote hydrogen- passivated surface, HF/H2O vapor cleaning induces fluorine terminated surface.
When the surface of wafer receiving HF/H2O vapor clean, the content of water
-
8/17/2019 Introduction to Microelectronics Fabrication
30/53
01 Introduction to Microelectronic Fabrication Preparation
- 26 -
vapor varies with the composition ratios, oxygen content can significantly affect
the content of fluorine. Increasing water content can reduce concentration of
fluorine.
Ar/H2 plasma cleaning is used to reduce the bombardment damage to the
surface of wafer. Ar and H2 gas molecules are excited or ionized and will
generate plasma with RF of 13.6Mz passing through induction coil at pressure
1.0torr. Excited Ar ion is physically sputtered the surface impurities away,while H2 ion chemically etch the surface. By proper adjustment of the physical
and chemical etching ratio, an optimum cleaning condition can be obtained that
produce minimum damaged surface.
1.3.3.3 Thermal Cleaning
The native oxide can be removed by heating the wafer to 800°C or above in
ultra high vacuum (< 10-10
torr) to vaporize the oxide. The native oxide is SiOx,
in which x depends on the previous cleaning process. In the Shiraki cleaning
process, temperature of 850°C and 10 min duration thermal clean is necessary.
However, in the HF dip process with 4% HF, a pre-bake at temperature 200°Cis required or no bake process if it is used for a 650°C epitaxial growth.
In all these processes, high temperature cleaning should be carefully
examined because at temperature higher than temperature 800°C, the following
reaction process occurs at low oxygen partial pressure.
Si +SiO2 → 2SiO (1.25)
The SiO is volatile at temperature above 750°C. When the SiO2 film is
removed, the silicon wafer starts to oxidized following equation (1.26).
Si +O2 → SiO2 (1.26)
This produces aggravated etching, which induces micro-roughness on the
surface. The deposited gate oxide, therefore, has a low breakdown voltage(~1.0MV/cm) due to micro-roughness.
1.3.4 Gettering
Gettering is the third line of control to avoid ionic contaminant reaching the
active region of the integrated circuit after facility cleaning and wafer cleaning.
The active regions of the integrated circuit usually occupy a small fraction of
-
8/17/2019 Introduction to Microelectronics Fabrication
31/53
01 Introduction to Microelectronic Fabrication
- 27 -
the volume of wafer. If there are contaminants resided on them, once they aredriven away, the performance of the circuit usually will not be affected because
the concentration of contaminant on the active region is now too low to be
influential.
Gettering is a process of moving the unwanted contaminant resided in thewafer to the non-critical part of wafer such as the backside of the wafer or far
away from the active parts on the top of wafer. The contaminants that are the
most concern which requiring gettering, are heavy transition elements such as
titanium, chromium, mercury, copper etc. They are normally the deep level
contaminants found in silicon wafer due to high diffusion coefficient. The other
most concern contaminants are alkali ions such as sodium Na+ and potassium
K
+
that usually come from human sweat commonly residing in dielectricmaterial that can cause threshold shift of the MOSFET.
The processes of gettering consist of three steps. Firstly, the elements to be
gettered must be freed from any trapping sites that they are currently occupiedand made mobile. Secondly, they must diffuse to the gettering site and finally,
they must be trapped permanently.
1.3.4.1 Alkali Gettering
Phosphosilicate glass PSG containing 5% by weight of phosphorus, is normally
deposited on top of the wafer to prevent contamination. It is a good material that
can drive alkali ions from contaminating the dielectric material or draw alkali
contaminant from dielectric material. PSG that has chemical name P2O2/SiO2, is
normally deposited using chemical vapor deposition CVD or low pressurechemical vapor deposition LPCVD covering the top of the wafer. It traps alkali
ions and forms a stable compound that binds sodium Na+ and potassium K +
ions. Thus, it is an effective way to prevent these ions diffusing into the
dielectric region or drawing these ions from dielectric materials.
The shortcoming of PSG is its charge dipole nature. After trapping ions, itcan affect the surface electric field. Moreover, it is susceptible to absorbing
water vapor that can cause aluminum corrosion. These effects can be minimized
by controlling the percentage of phosphorus content. In the case of MOSFET
fabrication, silicon nitride Si3 N4 is used to prevent this problem. This layer is
impermeable to alkali ions and can form effective barrier to diffusion.
-
8/17/2019 Introduction to Microelectronics Fabrication
32/53
01 Introduction to Microelectronic Fabrication Preparation
- 28 -
1.3.4.2 Heavy Metal Gettering
Heavy metallic elements such as copper, nickel etc have two basic properties.These elements have high diffusion coefficient. Thus, they are normally deeply
trapped in the silicon wafer. The second property is that atoms of these elements
have preferential residing site in the crystal lattice where imperfection exist.
Thus, one can see that the principle of gettering contamination of heavy metal isto create imperfection site in the crystal lattice or trapping center.
There are two basic gettering methods to draw away heavy metallic
element, which are extrinsic and intrinsic getterings. The extrinsic gettering
makes used of the backside of the wafer explicitly creating trapped center totrap such contaminant. Intrinsic gettering makes use of the properties ofCzochralski CZ silicon that contains oxygen to precipitate and trap metallic
contaminants because oxygen is a recombination center. Note that during the
Czochralski silicon ingot growth process, the molten silicon etches or dissolves
the quartz crucible wall to generate oxygen.
There are many methods to form extrinsic gettering which is normally
done at the backside of wafer. Methods are grinding, sand paper, ion
implantation, laser melting, depositing amorphous or polycrystalline silicon etc.
All of these methods are meant to create crystal lattice imperfection or extend
dislocation so that the metallic ion can be occupied during subsequently hightemperature process due to its high diffusion coefficient.
Intrinsic gettering uses the oxygen content in CZ silicon, which is normally
at the substrate region of the silicon wafer distanced away from the epitaxial
layer which is part of the device region. The concentration of oxygen in the CZsilicon is normally in 10 to 20ppm, which is about 1018cm-3. Oxygen can react
with silicon to form silicon dioxide SiO2 due to high process temperature.
Owing to the difference in size of SiO2 and silicon atom, stack fault is normallyoccurred which means dislocation or imperfection exist for precipitation or trap
site for the heavy metallic contaminant. One may ask why precipitation does notoccur in epitaxial layer. The answer is that the epitaxial layer is a fabrication
process that it has much lower oxygen content and is not sufficient enough for
precipitation.
1.4 Crystal Growth
-
8/17/2019 Introduction to Microelectronics Fabrication
33/53
-
8/17/2019 Introduction to Microelectronics Fabrication
34/53
01 Introduction to Microelectronic Fabrication Preparation
- 30 -
form of high purity. This is the raw material used to prepare device quality
single crystal. Pure EGS has impurity concentration generally in part per billion.
The pure EGS is then ready to be pulled into silicon ingot for making wafer that
used to fabricate integrated circuit.
There are a number of methods used to grow silicon crystalline ingot. We
shall discuss three methods here namely Czochralski, Float-zone methods, and
Bridgman-Stockbarger technique. There are other methods such as liquidencapsulated Czochralski LEC, micro-pulling-down µ-PD, laser-heated pedestal
growth LHPG or laser floating zone LFZ etc are not discussed here.
1.4.1 Czochralski Crystal Growth Method
Czochralski crystal growth method was invented by a polish chemist named JanCzochralski. The polycrystalline silicon is melt at temperature 1,4150C just
above the melting point temperature of silicon, which is 1,4140C, in the argon
Ar atmosphere in quartz crucible by radio frequency RF or resistive heating
coil. Right type and the amount of dopant are then added. With the aid of
“seed”, silicon rod of right diameter is formed by rotation and pulling in
Czochralski CZ puller as shown in Fig. 1.16. Figure 1.16(a) shows the photograph of a modern computer-controlled Czochralski crystal puller. Figure
1.16(b) is the schematic drawing showing the components of the puller.
Once thermal equilibrium is established, the temperature at the vicinity of
the seed is reduced and the molten silicon begins to freeze out onto the seedcrystal. Subsequently, the seed is slowly rotated and withdrawn at the rate of a
few millimeter per minute to form a cylindrically shaped single crystal of
silicon, which is known as ingot .
The diameter of the crystal in CZ method can be controlled by temperature
and pulling rate using automatic diameter control system. Typically, 4 to 6 inchdiameter and 1 to 2 meter in length type of ingot can be formed. In today’s
process, ingot of diameter as large as 12 inches is commonly produced to save
cost and improve productivity. However, for large ingot as large as 12 inches in
diameter, an external magnetic field is applied around the crucible and it is usedto control the concentration of defects, impurities, and oxygen.
18 inches wafer is currently under studied and it is expected to turn out first
wafer in 2018.
-
8/17/2019 Introduction to Microelectronics Fabrication
35/53
01 Introduction to Microelectronic Fabrication
- 31 -
(a) Physical appearance (b) Internal structureFigure 1.16: A Czochralski silicon rod puller
1.4.1.1 Impurity of Czochralski Process
The crystal ingot growth by Czochralski method always has trace impurities of
oxygen and carbon, which come from silica and graphite crucible materials.
Silica is silicon dioxide, which is the source of oxygen. Silica will react with
graphite to form carbon monoxide, which is the source of carbon. The equation
of chemical reaction is shown in equation (1.28).
SiO2 + 3C SiC + 2CO (1.28)
Typically, the oxygen concentration is approximately ranged from 1.0x1016cm-3
to 1.5x1018
cm-3
and the concentration of carbon varies from 2.0x1016
cm-3
to
1.0x1017
cm-3
. The contents of oxygen and carbon are very much depending on
ambient pressure, pulling and rotation rate, and the ratio of the diameter and thelength of the ingot.
-
8/17/2019 Introduction to Microelectronics Fabrication
36/53
01 Introduction to Microelectronic Fabrication Preparation
- 32 -
1.4.1.2 Concentration of Czochralski Process
In the crystal growth process, the most common dopants is boron and
phosphorus, which are used to make p- and n-type semiconductor materialsrespectively. As the crystal is pulled from the molten silicon, the doping
concentration incorporated into the crystal is usually different from the doping
concentration of the molten silicon at the interface. The ratio of these two
concentrations is equilibrium segregation coefficient k 0, which is defined as
k 0 I
S
C
C (1.31)
where Cs and CI are respectively the equilibrium concentration of the dopant inthe solid and liquid near interface. Figure 1.17 shows the equilibrium
segregation coefficient of the common dopants used for silicon. The value
below one means that during the growth the dopants are rejected into the moltensilicon. As the result, the dopant concentration of molten silicon becomes higher
as time lapsed.
Dopant k o Type Dopant k o Type
B 8.0x10-1
p As 3.0x10-1
n
Al 3.0x10-3
p Sb 3.3x10-2
nGa 8.0x10
- p Te 3.0x10
- n
In 4.0x10-4 p Li 1.0x10-2
Deep-lying
impurity
level
O 1.25 n Cu 4.0x10-1
Deep-lyingimpurity
level
C 7.0x10
-2
n Au 5.0x10
-5
Deep-lying
impuritylevel
P 0.35 nFigure 1.17: Equilibrium segregation coefficients for dopant in silicon
Let’s consider a crystal being growth from the initial molten silicon of wei ghtMo with an initial doping concentration Co (the weight of dopant per 1g of
molten silicon) in the molten silicon. At a given time, a crystal of weight M has
been grown, the amount of the dopant remaining in the molten silicon by weight
is S. For an incremental amount of the crystal with weight dM, the
-
8/17/2019 Introduction to Microelectronics Fabrication
37/53
01 Introduction to Microelectronic Fabrication
- 33 -
corresponding reduction of the dopant -dS from the molten is CSdM, where CS is the doping concentration in the crystal by weight.
-dS = CSdM (1.32)
The remaining weight of the molten silicon is (Mo – M) and the dopingconcentration in liquid by weight CI is given by
CI =MM
S
o (1.33)
Substituting equation (1.32) and (1.33) into equation (1.31), it yields equation
(1.28).
MM
dMk
S
dS
o
o (1.34)
Given that the initial weight of the dopant is CoMo, integration equation (1.34)
yields equation (1.35).
M
0 oo
S
MC MM
dMk
S
dS
oo
(1.35)
Solving equation (1.35) and combining with equation (1.33), it yields equation
(1.36).
1k
o
ooS
0
M
M1Ck C
(1.36)
During the growth of silicon ingot, dopant is constantly being rejected into the
molten silicon. If the rejection rate is higher than the rate at which the dopant
can be transported away by diffusion or stirring, then a concentration gradient
will develop at the interface as shown in Fig. 1.18. The equilibrium segregationcoefficient is equal to k 0 = CS/CI(0). We can define an effective segregationcoefficient k e, which is the ratio of CS and the impurity concentration far away
from the interface.
k e I
S
C
C (1.37)
-
8/17/2019 Introduction to Microelectronics Fabrication
38/53
01 Introduction to Microelectronic Fabrication Preparation
- 34 -
Figure 1.18: Doping distribution near the solid-molten interface
Let’s consider a small virtual stagnant molten layer of width in which the only
flow that required to replace the crystal being withdrawn from the molten.
Outside the stagnant layer the concentration remains constant at CI. In the layer,
the concentration can be described by steady state continuation equation.
0dx
dCvdx
CdD2
2 (1.38)
where D is the diffusion coefficient of the molten silicon and v is the velocity of
the crystal growth. The solution of equation is C = 2D/vx
1 AeA with the
constant to be determined by two boundary conditions. The first is at x = 0, C =
CI(0) and second is determined by conservation of total number of dopant i.e.
the sum of dopant flux at interface is zero. This condition yields equation.
0C)0(CdxdCD SI
0x
(1.39)
Substituting the conditions and C = CI at x = , the solution for the
concentration C is
SI
SID/V
C)0(C
CCe
(1.40)
The effective segregation coefficient k e
-
8/17/2019 Introduction to Microelectronics Fabrication
39/53
01 Introduction to Microelectronic Fabrication
- 35 -
D/v
00
0
I
Se
e)k 1(k
k
C
Ck
(1.41)
1.4.1.3 Pull Rate of Czochralski Process
Pertaining pull rate of Czochralski crystal growth, one expects the pull rate should
be slower for larger diameter ingot. Indeed the pull rate is inversely proportional
to the square root of diameter of ingot. It can be derived based on the first order
heat balance equation, which represents the dominant heat fluxes present during
freezing process. Reference to Fig. 1.19, x1 is a constant temperature surface,which is isotherm just inside the liquid. x2 is an isotherm just inside the solid.
During freezing process, which occurs between these isotherms, heat is released
to allow the silicon to transform from liquid to solid state, which is heat offusion. This heat must be removed from freezing interface. It is a primary
process of heat transfer by conduction up to the solid ingot. Thus, one can writeequation (1.42).
2
2
S1
1
L Adx
dTk A
dx
dTk
dt
dmL (1.42)
where L is the latent heat of fusion, dm/dt is the amount of silicon freezing per
unit time, k L is the thermal conductivity of liquid, dT/dx1 is the temperaturegradient across the isotherm x1, k s is the thermal conductivity of the solid.
dT/dx2 is the temperature gradient across the isotherm x2, and A1 and A2 arerespectively the cross sectional areas.
Figure 1.19: Freezing process occurring during Czochralski crystal growth
-
8/17/2019 Introduction to Microelectronics Fabrication
40/53
01 Introduction to Microelectronic Fabrication Preparation
- 36 -
The middle term of equation (1.42) will drop from this point onward, which is
representing any additional heat may flow from the liquid to the solid because
of the temperature gradient between the two. By neglecting it, one can include
only the absolute minimum heat which must be transported away from thefreezing interface. The effect on the final result will be that the pull rate will be
the maximum. If area A1 and A2 is equal to A then the rate of crystal V p is
pulled out of the molten silicon is simply equal to
ANVdt
dm p (1.43)
Substituting equation (1.43) into equation (1.42) after ignoring the middle term
of the equation, it becomes
2
s
MAX p dx
dT
LN
k V (1.44)
V pMAX is the maximum pull rate and N is the density of silicon.
In order to eliminate the temperature gradient term from equation (1.44),
one needs to consider how the heat is conducted up the solid crystal and how it
is eliminated from the solid ingot. Reference to Fig. 1.19, the latent heat ofcrystallization (A) is transferred from molten silicon to solid ingot. The heat is
then transported away from the freezing interface primarily by conduction up
the solid ingot (B). The heat is lost from ingot by radiation (C) and by
convection, although one will consider only radiation to keep the analysissimple.
The Stefan-Boltzmann law describes heat loss due to radiation (C) is
)T)(rdx2(dQ 4 (1.45)
where 2rdx represents the radiating surface area of an increment length of the
ingot. is Stefan-Boltzmann constant and is the emissivity of the silicon.
The heat conducted up the ingot (B) is given by
dx
dTr k Q
2
S (1.46)
-
8/17/2019 Introduction to Microelectronics Fabrication
41/53
-
8/17/2019 Introduction to Microelectronics Fabrication
42/53
01 Introduction to Microelectronic Fabrication Preparation
- 38 -
Equation (1.52) has clearly shown that the maximum pull rate V pMAX is
proportional to square root of the ingot’s radius.
1.4.2 Float-Zone Crystal Growth Method
The method is discovered by William Gardner Pfann. The float-zone crystal
growth method is illustrated conceptually in Fig. 1.20. The crystal is not grown
in the crucible that it has markedly reduced the impurity level particularly the
level of oxygen and carbon. It is grown in sealed furnace with argon Ar gas.
This method is used today for fabricating device that requires high resistivityand low oxygen content in the power device and detector device.
In the float-zone process, a polysilicon rod of EGS is clamped at both ends,with bottom in contact with a single-crystal seed. A small RF coil provides largecurrent in silicon that locally melts the silicon. The molten zone is usually
2.0cm long. The liquid phase silicon is then bonded to the atomic plane of the
seed plane by plane as the zone is slowly moved up. Doping of the crystal can
be achieved by either starting with a doped polysilicon rod, a doped seed, or
maintaining a gas ambient during the process that contains a dilute
concentration of the desired dopant.
Figure 1.20: Basic float-zone crystal growth method
-
8/17/2019 Introduction to Microelectronics Fabrication
43/53
01 Introduction to Microelectronic Fabrication
- 39 -
Segregation effect also plays an important role in the float-zone process just asit did in Czochralski method. It is illustrated from derivation of concentration of
solid silicon Cs(x) formed as it moves from molten state at the bottom to the top.
Figure 1.21 shows the idealized geometry of zone length L. The rod has
initial concentration of C0.
If the molten zone moves upwards by a distance dx, the number of
impurities in the liquid zone will change since some will dissolve into the
melting liquid at the top and some will be lost to the freezing solid at the
bottom. Thus,
dx)Ck C(dI I00 (1.53)
where I is the number of impurities in the liquid. However, concentration of
molten silicon is CI = I/L. Thus, substituting it into equation (1.53) andintegrating, it yields equation (1.54).
I
I 00
x
0 0
L
Ik C
dIdx (1.54)
where I0 is the number of impurities in the zone when it is first formed at the bottom. Performing the integration and noting that I0 = C0/L and Cs = k 0I/L, the
concentration of solid Cs(x) at distance x is
L
xk
00S
0
ek 11C)x(C (1.55)
As compared to Czochralski method, float-zone method has a greater resistivity
variation. Thus, Czochralski method is still the dominant method for largediameter silicon crystal. The melt-crystal interface is very complex for float-
zone method, so it is difficult to get dislocation free crystal. Unlike Czochralski
method, it needs a high-purity polysilicon to begin.
-
8/17/2019 Introduction to Microelectronics Fabrication
44/53
01 Introduction to Microelectronic Fabrication Preparation
- 40 -
Figure 1.21: Float-zone crystal growth process from liquid zone at the bottom moving to top
1.4.3 Bridgman-Stockbarger Crystal Growth Method
The set-up of Bridgman-Stockbarger crystal growth technique is shown in Fig.
1.22. This method is named after after Harvard physicist Percy Williams
Bridgman and MIT physicist Donald C. Stockbarger. The molten silicon or
germanium is placed in crucible. With a small crystal placed at one end. Thecrucible is pulled slowly in the horizontal direction. As the molten zone moves
out of the furnace, it is slowly cooled and silicon or germanium solidifies
following the crystal.
Figure 1.22: Bridgman-Stockbarger crystal growth method
1.5 Wafer Process
After the growth of silicon ingot, the ingot is machined cut at the end and
polishes the sides to remove grooves created by automatic diameter control
-
8/17/2019 Introduction to Microelectronics Fabrication
45/53
01 Introduction to Microelectronic Fabrication
- 41 -
system. The flat of 150mm or smaller, or notch of 200mm or larger is grindedon the ingot to mark the crystal orientation.
The ingot is then sliced into wafer thickness of about 150µm depending onthe diameter of the wafer using a rapid-rotating inward-diameter diamond-
coated saw. After sawing, the wafer is grinded to round shaped edge to preventedge chipping during mechanical handling of wafer processing.
The wafer is then rough polished by conventional abrasive, glycerin slurry
with fine alumina Al2O3 suspension-lapping process to remove the majority of
surface damage. It is necessary to create a flat surface for photolithography. The
process can produce surface flatness within 2.0m. During the lapping process,
about 50m of silicon is removed from both side of 150µm wafer.
After lapping process, a wet etch process is necessary to remove remaining
damage that can be as deep as 10m into the silicon. Thus, wet etch would
remove 10m from both sides of the wafer. The wet etchant is a mixture of
nitric acid HNO3, hydrofluoric acid HF, and acetic acid CH3COOH. Nitric acid
reacts with silicon to form silicon dioxide, while hydrofluoric acid then removes
the oxide. Acetic acid is used to control the rate of reaction. The wet process
further smoothen the surface due to isotropic etch characteristics with nitric
acid-rich solution. The usual formation is a 4:1:3 mixture of nitric acid (79wt%in H2O), HF (49wt% in H2O), and pure acetic acid. The chemical reaction
equation for the wet etch process is expressed in equation (1.56).
3Si + 4HNO3 + 6HF → 3H2SiF6 + 8H2O + 4NO↑ (1.56)
The last process is chemical mechanical polishing CMP as shown in Fig. 1.23.
Chemical mechanical polishing process has been used to prepare silicon Si
wafers for more than 30 years, as well as for glass polishing and for bonded
silicon-on-insulator SOI wafers. Although its application to interlayer dielectric
ILD planarization is more recent, both the equipment and the technique are wellknown to the semiconductor industry. The wafer is held on a rotating holder and
pressed on a rotating polishing pad with slurry and water in between. The slurry
is a colloidal suspension of fine silica SiO2 particle with diameter of about
100o
A in an aqueous solution of sodium hydroxide. Sodium hydroxide oxidizes,
which is a chemical process, the silicon surface with the help of heat generated
by friction between wafer and polishing pad. The silica particle then abrades the
silicon dioxide away from the surface – a mechanical process.
-
8/17/2019 Introduction to Microelectronics Fabrication
46/53
-
8/17/2019 Introduction to Microelectronics Fabrication
47/53
01 Introduction to Microelectronic Fabrication
- 43 -
mentioned in earlier Section. Back side defect can also be created using argonion implantation, polysilicon deposition, and heavily doped phosphorus. During
fabrication, the backside of the wafer is usually always deposited with a
chemical vapor deposition CVD silicon dioxide or silicon nitride Si3 N4 layer to prevent any out diffusion during thermal process.
Exercises
1.1. Name two factors that make silicon the most attractive semiconductor
material in electronic application.
1.2.
Name three properties and applications of wide band-gap semiconductor
materials.1.3. Explain why the conductivity of the intrinsic semiconductor is low at
room temperature.
1.4.
A silicon step junction at temperature 300K equilibrium condition has a
p-side doping with NA = 2x1015
cm-3
and n-side doping with ND =
1x1017
cm-3
. Calculate the built-in potential V bi.
1.5.
Determine all carrier concentrations in a Si p+n junction and given that
the slope of 1/C
2
versus VR graph is 1.32x10
14
Fcm
-2
V
-1
and built-in potential V bi is 0.850V.
1.6. Discuss conceptually how an npn Si bipolar junction transistor shall bedesigned?
1.7. What are carrier components in an npn Si bipolar junction transistor when
it is biased in forward active mode?
1.8. Define the emitter efficiency e for a BJT. Discuss how you can improve
the emitter efficiency for a bipolar junction transistor.
1.7.
Consider an n-channel MOSFET with gate width W = 10m and gate
length L = 1.5m and oxide capacitance Cox = 10-7F/cm2. In the linear
region for a fixed VDS = 0.1V, the drain current is found to be 40A for
VGS = 1.5V a
top related